W3C home > Mailing lists > Public > www-ws-desc@w3.org > April 2003

RE: WS-Addressing and R085

From: Clemens F. Vasters <clemensv@newtelligence.com>
Date: Sun, 13 Apr 2003 09:19:25 +0200
Message-ID: <20A058F869F7BC44B0E105207593C6BD0FA508@voyager.newtelligence.com>
To: "Mark Baker" <distobj@acm.org>, <www-ws-desc@w3.org>

Pardon me for jumping in. Usually I am a "silent reader" here, but I have a problem with "protocol independence is a bug".
 
(a) I just wrote yet another mail to a customer who needs to broadcast to multiple (thousands) of web service endpoints via UDP datagrams sent over a satellite feed.
(b) I know of an Asian stock exchange which employs a queueing network connecting autonomous computing agents for handling their trading system, for which any transport expect message queues would be inacceptable at present -- still the payload they carry are WS calls.
(c) Last week I talked to a medical imaging company who is using plain TCP connections carrying DIME streams to send large sequences of image data tagged with metadata enclosed in SOAP envelopes.
 
All different protocols, all proper uses of web service technology (and one of them actually using WS-Addressing already)
 
WS-Addressing addresses what I consider to be the greatest of all sins in WSDL and that's mixing message contract with binding to a concrete endpoint and concrete protocol (<service>). 
 
WS-Addressing is a separate construct to express endpoints. Many WS-Addressing instances can apply to a single WSDL -- which makes perfect sense in the presence of routing. If you use routing, any binding in the WSDL must even be ignored, because it usually doesn't tell you anything about the endpoint you need to talk to next.
 
Best Regards,
Clemens Vasters
 
[I could, but I am not even getting into the additional overlaps with WS-Policy at this time]

	-----Original Message----- 
	From: Mark Baker [mailto:distobj@acm.org] 
	Sent: Sun 4/13/2003 05:28 
	To: www-ws-desc@w3.org 
	Cc: 
	Subject: Re: WS-Addressing and R085
	
	


	On Fri, Apr 11, 2003 at 01:02:26PM -0400, Arthur Ryman wrote:
	> The WS-Addressing spec has recently been published [1]. It describes how
	> to send Web service endpoint references in messages and is therefore very
	> relevant to R085. Pls review and comment. Thx.
	
	Well, I think you said it there when you said that WS-Addressing deals with
	endpoint references.  R085 appears to require that WSDL be able to describe
	messages with URIs anywhere in the message, pointing to anything (e.g.
	perhaps those involved in a hypermedia interaction), not just endpoint
	references in a particular portion of the message (e.g. headers).
	
	I think R085 requires a much more general solution than WS-Addressing
	provides.
	
	And FWIW, I don't think WS-Addressing is a solution to any known
	problem.  It assumes "protocol independance" is a feature, rather than
	a bug.
	
	MB
	--
	Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
	Web architecture consulting, technical reports, evaluation & analysis
	
	
Received on Sunday, 13 April 2003 03:24:45 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:23 GMT